Chipping away at Power Module Temperatures

August 1, 2016

Terry Costlow

If power modules use silicon carbide devices, thermal management can be simpler, according to Wolfspeed’s Kierstead.

Semiconductors are tiny compared to the off-highway vehicles that carry them, but their thermal issues aren’t small. Keeping the sub-micron elements of a chip cool is a major challenge for the developers of electronic controls.

In today’s mobile world, chips are increasingly designed for lower power and cool operations. Nonetheless, harsh off-highway environments force design teams to take extra care to use semiconductors that add very localized heat to the ambient temperatures found in engine compartments and other hot, cramped areas. In these dirty environments, passive cooling is always preferable to fans. That means chips have to offer power and heat specifications that fit their surroundings. Placing controls as far away from hot spots is critical.

“This is achieved through careful selection of electronic components and consideration of thermal conditions in design of the enclosure for the electronics,” said Shelley Knust, Executive Director, Engine Business Unit Off-Highway Engineering at Cummins. “This also requires knowledge of the ambient conditions of the intended mounting location of the controller.”

Cooling the semiconductors in controllers comes with many trade-offs. In many modules, CMOS devices are the obvious choice. But in modules that work with high power levels, CMOS may not always provide the best thermal traits.

When modules manage electric motors and other more demanding equipment, alternative semiconductor technologies can often meet performance requirements while providing more flexibility to meet electrical or thermal demands.

“Silicon carbide can cut switching losses in half,” said Paul Kierstead, Director of Marketing for Wolfspeed's Power Division, which is owned by Cree Inc. “You can use that to improve efficiency or you can move to higher frequency conversion electronics. Higher frequencies mean you can use smaller inductors and other components. This significantly reduces cooling requirements, so instead of using liquid cooling for silicon devices, you can use convection cooling for silicon carbide.”

Regardless of the semiconductors being used, convection cooling is generally the desired methodology. Liquid or fan-based cooling systems add complexity and often bring reliability issues.

“Fans would require controllers for these applications to reside either on-engine or in the engine compartment,” Knust said. “Environmental conditions and normal vehicle maintenance procedures, such as power washing of the engine compartment, would damage cooling fans. Fluid cooling (whether fuel or engine coolant) is not preferred due to the complexity of installation during both manufacturing and service.”

While cooling is always a part of any electronic design, its impact doesn’t usually extend beyond the module level. When engineers are creating architectures with centralized and distributed intelligence, thermal management isn’t a consideration.

“The impact of cooling is not a primary factor in our electronic control system architectures including whether modules are consolidated or distributed,” Knust said.